Highliner

Installation:

To install required libraries: pip install -r requirements

Inline instruction identification

usage: highliner.py [-h] [-o OUTPUT_FILE] [-b BINARY_PATH] [-nogpu] [-t THRESHOLD] input_file

input file is expected to be the JSON file produced by Bino by running it with the -o [OUTPUT_FILE] option

Optional parameters:

• -o: output file to dump the results in as json.
• -b: binary file path which Bino analyzed (default: get from Bino output)
• -t: threshold to distinguish between positive and negative class. Should be between 0 and 1. (default: optimal testing threshold)
• -nogpu: disables GPU usage

Same binary test sample as BINO available in data/test_binaries/generic_binaries/, to have the tool still work with default options

Output file will be the same input dictionary with matches extended to include instructions and predicted probability of them being inline

Dataset

Can be downloaded at: https://drive.google.com/file/d/17TLq5zP-2nKjNDEOC4rIw26eUv_LB7OJ/view?usp=sharing